Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

VQA

Benchmarks

Task NameDataset NameSOTA ResultTrend
Visual Question AnsweringVQA v2
Accuracy100
1,429
Visual Question AnsweringVQA v2 (test-dev)
Overall Accuracy87.66
712
Visual Question AnsweringVQA v2 (test-std)
Accuracy86.1
486
Visual Question AnsweringVQA 2.0 (test-dev)
Accuracy86.5
337
Visual Question AnsweringVQA v2
Accuracy88.5
333
Visual Question AnsweringVQAv2
Accuracy97.4
196
Visual Question AnsweringVQA 2.0 (val)
Accuracy (Overall)86.1
183
Visual Question AnsweringVQA v2 (val)
Accuracy95.06
158
Visual Question AnsweringVQA (test-dev)
Acc (All)78.25
147
Visual Question AnsweringVQA v2 (test)
Accuracy86.1
142
Visual Question AnsweringVQA (test-std)
Accuracy84
120
Open-Ended Visual Question AnsweringVQA 1.0 (test-dev)
Overall Accuracy66.7
100
Visual Question AnsweringVQAv2 (test)
VQA Accuracy83.4
82
Visual Question AnsweringVQAv2 (test-dev)
Accuracy86.1
80
Visual Question AnsweringVQA v2
Accuracy80.1
71
Visual Question AnsweringVQA
Accuracy82.18
66
Visual Question Answering (Multiple-choice)VQA 1.0 (test-dev)
Accuracy (All)70.04
66
Visual Question AnsweringVQA text
Accuracy83.2
61
Visual Question AnsweringVQA (val)
Overall Accuracy79.54
55
Visual Question AnsweringVQA v2
VQAv2 Accuracy81.8
50
Open-Ended Visual Question AnsweringVQA 1.0 (test-standard)
Overall Accuracy67.36
50
Visual Question AnsweringVQA v2
Overall Accuracy89.4
45
Visual Question AnsweringVQAv2 (test-std)
Accuracy82.3
44
Visual Question AnswerVQA 1.0 (test-dev)
Overall Accuracy67.42
44
Visual Question AnsweringVQA v2
ASR100
42
Showing 25 of 149 rows